Snapshots of two-dimensional jet flow in a tokamak divertor simulation, showing density, temperature, pressure, velocity, and vorticity contours. See page 64 for details.

There was not much to see five years ago when I arrived at the newly relocated NERSC in Berkeley — only a handful of new employees; anxious Berkeley Lab and DOE staff worrying whether their boldness would really pay off; major construction on the first floor of Building 50, which as yet bore no resemblance to a computer room; and trenches filled with rainwater in the parking lot. It seems remarkable how the pioneers of the new NERSC were undaunted by those challenges, and how, with neither staff nor computers in place, we boldly proclaimed that we would reinvent the high performance computing center. And, by all accounts, we did it.

By 1998 it became clear that the space and power requirements of the next few generations of high performance computing platforms would require a much larger computer room. Our Berkeley Lab colleagues in Operations and Facilities located and remodeled a cost-effective new facility in record time. In the fall of 2000, we moved most of our high-performance computing platforms to the new Oakland Scientific Facility, and our newest platform was installed in the first week of 2001, all on schedule.

The new platform will provide an unprecedented 3.8 Tflop/s peak performance for the DOE Office of Science computational community. While many of our colleagues at other sites brag about terascale computing, I believe that NERSC will be the first site where users can expect to see their applications routinely perform at teraflop/s level. This will be the culmination of years of work at NERSC to improve the utilization of highly parallel platforms, and to provide the tools that allow efficient execution of jobs requiring 512 processors and more. At the same time, our storage capability is nearing the petabyte level, thanks to continuing improvements over the last few years.

While integrating these new technologies, we continued to maintain the highest standards of service, and again enabled our community of users to attain breakthrough scientific results. The most notable accomplishment, among the many documented in this annual report, is a cover story in Nature, backed by data analysis carried out at NERSC.

In November, at the SC2000 conference in Dallas, Berkeley Lab and NERSC released three software CDs: Berkeley Lab AMR, Akenti, and Berkeley Lab VIA Software (M-VIA and MVICH). Highlights of these projects and many of our other R&D efforts are presented in this annual report, demonstrating the benefits of combining a computing facility with research and development in one organization.

After several years of planning, computational science in the DOE Office of Science finally received a big boost through the funding of the Scientific Discovery through Advanced Computing program (SciDAC). For NERSC’s clients, SciDAC offers a once-in-a-decade opportunity to demonstrate the applicability of past research and to engage in bold new projects. NERSC itself has the opportunity to further develop and then deploy the results of the computer science research of the last few years, enabling a whole new generation of computational science.

	Horst D. Simon, Division Director of NERSC.

I am pleased that the Office of Advanced Scientific Computing Research is re-examining its portfolio of research activities, and that NERSC will have the opportunity to develop a new five-year plan. Given our accomplishments of the last five years, I have no doubt that we will develop a first-rate strategy for NERSC and computational science in DOE, and that we will reinvent the high performance computing center yet again.

With these exciting times ahead of us, I am grateful to our DOE Office of Science sponsors for their continued endorsement of our ambitious plans. It continues to be a pleasure to collaborate with the NERSC Users Group and its executive board members. I would like to thank them for their continued support, especially for their effort to produce the next “Greenbook” documenting the computational requirements of the Office of Science community. My special thanks and congratulations, as always, go to the NERSC staff for their skill, dedication, and tireless efforts to make NERSC the best scientific computing resource in the world.